Calculation unloading and resource allocation method under relay-assisted MEC architecture

Abstract

The invention discloses a computing unloading and resource allocation method under a relay-assisted MEC architecture, and belongs to the technical field of wireless communication and computing resource management. The invention provides a joint optimization method of a task division strategy, a computing resource allocation strategy, a relay selection strategy and an uplink spectrum allocation strategy by introducing relay equipment and providing a plurality of task processing modes, including local execution, relay unloading and edge computing, in order to reduce the execution cost to the greatest extent from the perspective of a user, and provides a joint optimization method of the task division strategy, the computing resource allocation strategy, the relay selection strategy and the uplink spectrum allocation strategy. The network congestion can be reduced and the user experience can be improved while the service quality is ensured; in order to solve a joint optimization problem, block coordinate descent, a reconstruction linearization technology and a convex optimization method are utilized, the convergence speed is high, the calculation complexity is low, the relay-assisted MEC architecture and the optimization method can adapt to different types of intelligent devices and application scenes, wide applicability and flexibility are achieved, and the method has good application prospects. And the requirement of future mobile edge computing development is met.

Description

Technical Field

[0001] The present invention relates to a method for computing offloading and resource allocation under a relay-assisted MEC architecture, and belongs to the technical field of wireless communication and computing resource management. Background Art

[0002] The popularity of smart devices (SD) in modern communications and computing has led to a sharp increase in demand for computing power and storage capacity. However, due to hardware limitations, a single smart device cannot efficiently handle complex computing tasks. Therefore, Mobile Edge Computing (MEC) has emerged as a key technology to solve this problem, helping smart devices complete computing tasks by deploying computing resources at the edge of the network.

[0003] The introduction of relay devices in the MEC architecture can effectively make up for the lack of resources in traditional MEC systems. Relay devices can not only provide communication services, but also assist in completing computing tasks, thereby reducing the computing burden of smart devices and improving the computing power and resource utilization efficiency of the overall system.

[0004] In traditional MEC systems, resource management solutions usually only consider a single computing offloading mode and fail to fully utilize various resources in the network, resulting in low resource utilization, severe network congestion, and affected quality of service (QoS). Therefore, designing an optimization solution that can fully utilize multiple resources is an urgent problem to be solved. Summary of the invention

[0005] In order to solve the problems of low resource utilization and high execution cost in the current mobile edge computing (MEC) system, the present invention provides a method for computing offloading and resource allocation under a relay-assisted MEC architecture. The technical solution is as follows:

[0006] A method for computing offloading and resource allocation under a relay-assisted MEC architecture of the present invention includes:

[0007] Step 1: Build a relay-assisted MEC system model to define the functions and task processing modes of relay devices, smart devices, and edge servers;

[0008] Step 2: jointly formulate the task division strategy, computing resource allocation strategy, relay selection strategy and uplink spectrum allocation strategy in the relay-assisted MEC system model to form a joint optimization problem, wherein the goal of the joint optimization problem is to minimize the system execution cost while ensuring the service quality;

[0009] Step 3: Design a preliminary solution method, using the block coordinate descent method to decompose the joint optimization problem into multiple sub-problems, each of which optimizes the task division strategy, computing resource allocation strategy, relay selection strategy and uplink spectrum allocation strategy respectively;

[0010] Step 4: Further optimize the preliminary solution results, use the reconstruction linearization technique to approximate the non-convex problem as a convex optimization problem, and apply the convex optimization method to solve it.

[0011] Optionally, the step 1 includes:

[0012] Using the Decision Matrix To describe the relationship between the i-th smart device and the j-th relay device, where x i,j ∈{0,1} is a binary selection variable, x i,j =1 means that the i-th intelligent device selects the j-th relay device as a relay, and x i,j = 0 means that there is no matching relationship between the i-th smart device and the j-th relay device. According to the general physical meaning of relay selection, the variable x is selected. i,j The restrictions are as follows:

[0013]

[0014] Since the system adopts a partial offloading model, the task is divided into three parts: local computing, relay offloading, and edge offloading:

[0015]

[0016] in, and Describe the number of computing bits allocated to local computing, relay offloading, and edge offloading, respectively. i Indicates the total amount of data for this computing task;

[0017] For the i-th smart device, the latency is divided into three parts: local computing, relay offloading, and edge offloading. For relay devices and base stations, the offloading phase is divided into two states: transmission and calculation. The communication process is divided into two stages, corresponding to two transmission delays. In the first stage, the i-th smart device In the time slot The unloaded data is sent to the selected j-th relay device; the second stage is The jth relay device will receive Data is forwarded to the MEC server; use h i,j represents the power gain of the subchannel, B iis the subchannel bandwidth of the i-th smart device allocated to the j-th relay device, p i,j is the transmission power of the ith smart device, and N i,j is the noise power; if α i represents the ratio of the uplink transmission sub-bandwidth to the total bandwidth, then B i =Bα i ,α i ∈{0,1},∑ i α i ≤1;

[0018] In order to complete the assigned task within the maximum tolerable delay T, the time limit for local computing and relay offloading is:

[0019]

[0020] Among them C i is the number of CPU cycles required per bit, and Represents the CPU frequencies of the i-th smart device and the j-th relay device serving the assigned tasks, respectively satisfying and and They represent the maximum computing power that can be allocated to tasks by smart devices and relay devices, Indicates the delay of relay calculation;

[0021] The total energy consumption of the i-th smart device associated with the j-th relay device for local computing and task offloading is expressed as:

[0022]

[0023] in, is the power consumption per CPU cycle, κ i It depends on the chip architecture energy consumption coefficient, E comm represents the energy consumption of task offloading to the relay in the first transmission stage, E comp It represents the energy consumed by SD to complete local computing tasks.

[0024] Optionally, step 2 includes:

[0025] The unit cost of the relay receiving, storing and forwarding data supply to a smart device task is m r , calculate the supply price per bit of data as m c ; Use M i represents the service cost of supplying the i-th smart device task. When the task is successfully collected by the required relay, we get:

[0026]

[0027] For a given task partitioning strategy Computing resource allocation strategy Relay selection strategy and uplink spectrum allocation strategy The joint optimization problem is expressed as follows using the following formula (8):

[0028]

[0029] in, is the solution of P1, ω1 represents the unit cost of computing power service used by the i-th smart device, and ω2 is the weight of the offloading service Indicates the maximum CPU frequency used by the i-th smart device for local computing.

[0030] Optionally, step 3 includes:

[0031] Given a relay selection strategy and uplink spectrum allocation strategy Joint optimization task division strategy Computing resource allocation strategy Two sub-problems are expressed as:

[0032]

[0033] The objective function and constraints in subproblem P2 are both convex functions. Convex optimization techniques are used to solve P2 and obtain the local optimal solution under the current number of iterations.

[0034] Given a task division strategy Computing resource allocation strategy Jointly optimized relay selection strategy and uplink spectrum allocation strategy Two sub-problems are expressed as:

[0035]

[0036] in

[0037] Optionally, step 4 includes:

[0038] Using the slack variable method, the discrete variable x i,j Relax to 0≤x i,j ≤1;

[0039] make The value of θ is approximately 0;

[0040] For the second-order term Introducing auxiliary variables where xi,j and Respectively restricted to 0≤x i,j ≤1 and

[0041] Convert P3 to:

[0042]

[0043] Use convex optimization technology to solve P3.1, get the local optimal solution under the current number of iterations, and substitute the solution into And the optimization variable x is rounded off. i,j Perform recovery.

[0044] Optionally, orthogonal frequency division multiplexing is used for channel access in two stages: from the smart device to the relay device and from the relay device to the base station.

[0045] Optionally, the convex optimization method adopted in step 4 is a CVX solver.

[0046] Optionally, the task processing modes of step 1 include: local execution, relay offloading and edge computing.

[0047] The present invention provides an electronic device, comprising a memory and a processor;

[0048] The memory is used to store computer programs;

[0049] The processor is configured to implement any of the methods described above when executing the computer program.

[0050] The present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in any one of the above items is implemented.

[0051] The beneficial effects of the present invention are:

[0052] The present invention minimizes the user's overhead on communication and computing services by optimizing the computing offloading and resource allocation strategy, thereby reducing the overall execution cost of edge users. By introducing relay devices and reasonably selecting task processing modes (local execution, relay offloading, edge computing), efficient resource utilization is achieved and resource waste is avoided. By jointly optimizing computing offloading, relay selection, computing resource allocation and spectrum resource allocation, network congestion can be reduced and user experience can be improved while ensuring the quality of service (QoS). The two-stage iterative algorithm designed by the present invention utilizes block coordinate descent, reconstruction linearization technology (RLT) and convex optimization methods, which can effectively solve non-convex optimization problems, with faster convergence speed and lower computational complexity. The relay-assisted MEC architecture and optimization method can adapt to different types of smart devices and application scenarios, have wide applicability and flexibility, and meet the needs of future mobile edge computing development. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0054] Figure 1 It is a convergence performance curve diagram of the computation offloading and resource allocation method under the relay-assisted MEC architecture of the present invention.

[0055] Figure 2 It is a relationship diagram between user location distribution and relay selection in the task offloading scenario in the second embodiment of the present invention.

[0056] Figure 3 This is a schematic diagram comparing the total cost of the distance-based offloading scheme (D-JOA), the random offloading scheme (R-JOA), the sequence-based offloading scheme (O-JOA) and the technical solution of the present invention as the number of relay devices is 4 and the maximum tolerable delay is 1s, with the number of mobile devices varying.

[0057] Figure 4 It is a schematic diagram comparing the total cost of the above technical solutions with different maximum tolerable delays when the number of mobile devices is 20 and the number of relay devices is 4.

[0058] Figure 5 Based on the convex optimization technique, m r 1.5, m eWhen is 1, the present invention compares two benchmark algorithms, a MEC system with a single base station (SBS) and a MEC system assisted by D2D communication, to evaluate the overall execution cost curve under different numbers of relays.

[0059] Figure 6 This is a curve diagram of the total cost of the above technical solutions compared with different maximum tolerable delays in the present invention.

[0060] Figure 7 The task size is L i =200 bits, the relationship curve between the average execution cost, tolerable delay and the number of relays of the present invention.

[0061] Figure 8 It is a curve diagram showing the relationship between the task size and the number of relay devices when the maximum tolerable time T = 1s. DETAILED DESCRIPTION

[0062] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0063] Embodiment 1:

[0064] This embodiment provides a method for computing offloading and resource allocation under a relay-assisted MEC architecture, including:

[0065] Step 1: Build a relay-assisted MEC system model to define the functions and task processing modes of relay devices, smart devices, and edge servers;

[0066] Step 2: jointly formulate the task division strategy, computing resource allocation strategy, relay selection strategy and uplink spectrum allocation strategy in the relay-assisted MEC system model to form a joint optimization problem, wherein the goal of the joint optimization problem is to minimize the system execution cost while ensuring the service quality;

[0067] Step 3: Design a preliminary solution method, using the block coordinate descent method to decompose the joint optimization problem into multiple sub-problems, each of which optimizes the task division strategy, computing resource allocation strategy, relay selection strategy and uplink spectrum allocation strategy respectively;

[0068] Step 4: Further optimize the preliminary solution results, use the reconstruction linearization technique to approximate the non-convex problem as a convex optimization problem, and apply the convex optimization method to solve it.

[0069] Embodiment 2

[0070] This embodiment provides a method for computing offloading and resource allocation under a relay-assisted MEC architecture. The system model applicable to the method is: a relay-assisted MEC system, which consists of smart devices SDs, relay devices relays and a base station BS powered by an edge server. The smart devices and relay device sets in the system can be used separately. and Specifically, relay devices and base stations can provide communication and computing services for smart devices at the same time. Both have certain data storage and processing capabilities, which can assist smart devices in task forwarding and offloading.

[0071] There are three execution modes for tasks generated by applications, including local execution, relay offloading, and edge computing. In the relay device-MEC scenario, smart devices cannot communicate directly with base stations and must be forwarded through relay devices. In addition, in order to incentivize relay devices to serve more resource-hungry users and solve the problem of resource competition among users, this embodiment proposes a user charging system to sell the resources and services of relay devices and base stations to users through pricing. Each smart device will generate a random task that is computationally intensive and latency-sensitive, using data tuples. Indicates that C i Indicates the computing resources in terms of CPU cycles required to process a unit bit task, L i represents the total amount of data for this computing task, and T i Indicates the maximum tolerable delay for completing a task. In this embodiment, for simplicity, all tasks are assumed to be completed within a constant time block T.

[0072] Due to the limitations of link quality and computing resources, it is assumed that smart devices can be associated with relay devices one by one through some selection strategies. Due to the divisibility of tasks, these relay devices will help offload some tasks to MEC servers, and can also perform the remaining tasks within their capabilities. In this way, smart devices can offload tasks without consuming a lot of transmission energy. The quality of the relay device selection strategy directly affects the energy consumption of smart devices in the system, and also affects the cost of users obtaining resources.

[0073] The computation offloading and resource allocation method of this embodiment includes:

[0074] Step 1: Build a relay-assisted MEC system model and define the functions and task processing modes of relay devices, smart devices, and edge servers. The task processing modes include local execution, relay offloading, and edge computing.

[0075] Using a decision matrix To describe the relationship between the i-th smart device and the j-th relay device, where x i,j∈{0,1} is a binary selection variable, x i,j =1 means that the i-th intelligent device selects the j-th relay device as a relay, and x i,j = 0 means that there is no matching relationship between the i-th intelligent device and the j-th relay device. According to the general physical meaning of relay selection, the variable x is selected. i,j The restrictions are as follows

[0076]

[0077] Since the system adopts a partial offloading model, the task can be divided into three parts: local computing, relay offloading, and edge offloading, and the following relationship is satisfied between the three parts:

[0078]

[0079] in, and denote the number of computing bits allocated to local computing, relay offloading, and edge offloading, respectively, and L i Indicates the total data volume of this computing task.

[0080] For the i-th smart device, the latency can be divided into three parts: local computing, relay offloading, and edge offloading. In particular, for relay devices and base stations, the offloading phase is further divided into two states: transmission and calculation, that is,

[0081] The communication model is divided into two stages, corresponding to two types of transmission delays. In the first stage, the i-th smart device In the time slot The unloaded data is sent to the selected j-th relay device. Since the relay adopts a simplex communication mode, it needs to complete all data reception before performing calculation and forwarding tasks; The jth relay device will receive The process of forwarding data to the MEC server is in the second stage. The channel access from the smart device to the relay device and from the relay device to the base station adopts the orthogonal frequency division multiplexing (OFDM) method. i,j represents the power gain of the subchannel, B i is the subchannel bandwidth of the i-th smart device allocated to the j-th relay device, p i,j is the transmission power of the ith smart device, and N i,j is the noise power. If α i represents the ratio of the uplink transmission sub-bandwidth to the total bandwidth, then B i =Bα i ,α i ∈{0,1},∑i α i ≤1.

[0082] Since data cannot be directly offloaded from the local to the base station, the data that the relay needs to receive includes not only the data that it needs to execute itself, but also some data processed with the help of the base station computing power. In order to complete the assigned task within the maximum tolerable delay T, the time limit for the local calculation and relay offloading process is:

[0083]

[0084] in, represents the total latency of the local computing phase under the partial offloading model, represents the computation delay in the local computation phase, C i is the number of CPU cycles required per bit, and Represents the CPU frequencies of the i-th smart device and the j-th relay device serving the assigned tasks, respectively satisfying and Indicates the total delay of the relay unloading phase, represents the transmission delay in the relay offloading phase, Represents the total delay of the relay calculation stage, and They represent the maximum computing power that can be allocated to tasks by smart devices and relay devices respectively.

[0085] The total energy consumption of the i-th smart device associated with the j-th relay device for local computing and task offloading can be expressed as:

[0086]

[0087] Among them, E comm represents the energy consumption of task offloading to the relay in the first transmission stage, E comp It represents the energy consumed by SD to complete the local computing task. is the power consumption per CPU cycle, κ i It depends on the energy consumption coefficient of the chip architecture.

[0088] Step 2: Divide the Tasks Strategies Computing resource allocation strategy Relay selection strategy and uplink spectrum allocation strategy The problem is jointly formulated as a joint optimization problem (Joint Computation Offloading, Relay selection strategy, Computation Resource Allocation, and Spectrum Resource Allocation, JORA), and its goal is to minimize the system execution cost while ensuring the quality of service (QoS).

[0089] The unit cost of the relay receiving, storing and forwarding data supply to a smart device task is m r , and the supply price per bit of data is m c ; Use M i represents the service cost of supplying the i-th smart device task. When the task is successfully collected by the required relay, we can get:

[0090]

[0091] Considering the limitations of communication, computing power, etc., for a given task division strategy Computing resource allocation strategy Relay selection strategy and uplink spectrum allocation strategy The JORA optimization problem is expressed as follows using the following formula (8):

[0092]

[0093] in, is the solution of P1, ω1 can be understood as the unit cost of the i-th smart device using the computing power service, and ω2 is the weight of the offloading service to ensure fairness.

[0094] Step 3: Design a preliminary solution algorithm and use the block coordinate descent method to decompose the joint optimization problem (JORA) into multiple sub-problems, and optimize the task division strategy for each sub-problem Computing resource allocation strategy Relay selection strategy and uplink spectrum allocation strategy

[0095] Temporarily fix relay selection strategy and uplink spectrum allocation strategy Joint optimization task division strategy Computing resource allocation strategy Two sub-questions:

[0096]

[0097] The objective function and constraints in the subproblem are all convex functions, so P2 is a convex optimization problem. P2 can be solved by many convex optimization techniques, such as CVX, to obtain the local optimal solution under the current number of iterations.

[0098] For a given task partitioning strategy Computing resource allocation strategy Relay selection strategy for problem P1 and uplink spectrum allocation strategy It can be optimized by problem P3:

[0099]

[0100] in Obviously, problem P3 is non-convex due to the non-convex objective function.

[0101] Step 4: Use the reconstruction linearization technique (RLT) to approximate the non-convex problem into a convex optimization problem and apply the convex optimization method to solve it.

[0102] Considering the P3 relay selection strategy x i,j and spectrum allocation α i The coupling of the two terms is solved by first using the relaxation variable method to facilitate the design of low-complexity algorithms, and then using the constant addition method and reconstruction-linear technique (RLT) to eliminate the quadratic term. i,j is an integer variable from 0 to 1, which converts the discrete variable x i,j Relax to 0≤x i,j ≤1. At the same time, due to the possibility of division by zero, in the original variable α i Add a constant θ to the basis, let The value of θ is approximately 0. Since there is also a product term The transformation is performed using the RLT technique. For the second-order term Introducing auxiliary variables where x i,j and Respectively restricted to 0≤x i,j ≤1 and At this point, the transformed P3 can be formulated as:

[0103]

[0104] The transformed problem P3.1 is convex, so the solution to the problem can be obtained using many solvers (such as CVX). Substituting the solution into And the optimization variable x is rounded off. i,j Perform recovery.

[0105] The iterative optimization process is as follows:

[0106]

[0107] In order to make the purpose, technical solution and advantages of the present invention clearer, some classic detection algorithms will be used to compare with the proposed algorithm to show the performance superiority of the computational offloading and resource allocation method under the relay-assisted MEC architecture of the present invention. Figure 2 From the perspective of relay selection strategy, the relationship between user location distribution and relay selection in the task offloading scenario is presented.

[0108] The baseline algorithms used for simulation are distance-based unloading scheme (D-JOA), random unloading scheme (R-JOA) and order-based unloading scheme (O-JOA). Figure 3 , 4 As shown, it can be seen from the simulation curve that by comparing the total costs under different numbers of smart devices and different maximum tolerable delays, it is found that the computation offloading and resource allocation method of the present invention can effectively optimize the total system cost.

[0109] Two benchmark algorithms, a single base station (SBS) MEC system and a D2D communication-assisted MEC system, will be used to evaluate the performance of the JORA algorithm; at the same time, it uses convex optimization technology for iterative optimization. Figure 5 It shows that the JORA joint optimization method of the present invention has the best performance, and the overall execution cost increases with the increase in the number of relays. Figure 6 It is shown that as the maximum tolerable delay increases, the total system cost decreases, and the JORA joint optimization method of the present invention shows the most economical performance.

[0110] When the task size is set to L i =200 bits, Figure 7 The relationship between the average execution cost, tolerable delay, and the number of relays is shown. The relationship between the task size and the number of relays is shown in Figure 8 As shown, the maximum tolerable time T = 1 second.

[0111] In summary, the simulation results verify the performance of the JORA joint optimization method of the present invention in different scenarios, and compared with the baseline algorithm, the algorithm can greatly reduce the execution cost. Therefore, the computation offloading and resource allocation method of the present invention can effectively improve the performance of the network system while completing the SDs low-latency task.

[0112] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.

[0113] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for computing offloading and resource allocation under a relay-assisted MEC architecture, characterized in that: The method comprises: Step 1: Build a relay-assisted MEC system model to define the functions and task processing modes of relay devices, smart devices, and edge servers; Step 2: jointly formulate the task division strategy, computing resource allocation strategy, relay selection strategy and uplink spectrum allocation strategy in the relay-assisted MEC system model to form a joint optimization problem, wherein the goal of the joint optimization problem is to minimize the system execution cost while ensuring the service quality; Step 3: Design a preliminary solution method, using the block coordinate descent method to decompose the joint optimization problem into multiple sub-problems, each of which optimizes the task division strategy, computing resource allocation strategy, relay selection strategy and uplink spectrum allocation strategy respectively; Step 4: Further optimize the preliminary solution results, use the reconstruction linearization technique to approximate the non-convex problem as a convex optimization problem, and apply the convex optimization method to solve it.

2. The method according to claim 1, characterized in that The step 1 comprises: Using the Decision Matrix To describe the relationship between the i-th smart device and the j-th relay device, where x i,j ∈{0,1} is a binary selection variable, x i,j =1 means that the i-th intelligent device selects the j-th relay device as a relay, and x i,j = 0 means that there is no matching relationship between the i-th smart device and the j-th relay device. According to the general physical meaning of relay selection, the variable x is selected. i,j The restrictions are as follows: Since the system adopts a partial offloading model, the task is divided into three parts: local computing, relay offloading, and edge offloading: in, and Describe the number of computing bits allocated to local computing, relay offloading, and edge offloading, respectively. i Indicates the total amount of data for this computing task; For the i-th smart device, the latency is divided into three parts: local computing, relay offloading, and edge offloading. For relay devices and base stations, the offloading phase is divided into two states: transmission and calculation. The communication process is divided into two stages, corresponding to two transmission delays. The first stage is when the i-th smart device In the time slot The unloaded data is sent to the selected j-th relay device. The second stage is The jth relay device will receive Data is forwarded to the MEC server; use h i,j represents the power gain of the subchannel, B i is the subchannel bandwidth of the i-th smart device allocated to the j-th relay device, p i,j is the transmission power of the ith smart device, and N i,j is the noise power; if α i represents the ratio of the uplink transmission sub-bandwidth of the i-th smart device to the total bandwidth, then B i =Bα i ,α i ∈{0,1},∑ i α i ≤1, B represents the total bandwidth; In order to complete the assigned task within the maximum tolerable delay T, the time limit for local computing and relay offloading is: Among them C i is the number of CPU cycles required per bit, and Represents the CPU frequencies of the i-th smart device and the j-th relay device serving the assigned tasks, respectively satisfying and and They represent the maximum computing power that can be allocated to tasks by smart devices and relay devices, Indicates the delay of relay calculation; The total energy consumption of the i-th smart device associated with the j-th relay device for local computing and task offloading is expressed as: in, is the power consumption per CPU cycle, κ i It depends on the chip architecture energy consumption coefficient, E comm represents the energy consumption of task offloading to the relay in the first transmission stage, E comp It indicates the energy consumed by smart devices to complete local computing tasks.

3. The method according to claim 2, characterized in that The step 2 comprises: The unit cost of the relay receiving, storing and forwarding data supply to a smart device task is m r , calculate the supply price per bit of data as m c ; Use M i represents the service cost of supplying the i-th smart device task. When the task is successfully collected by the required relay, we get: For a given task partitioning strategy Computing resource allocation strategy Relay selection strategy and uplink spectrum allocation strategy The joint optimization problem is expressed as follows using the following formula (8): in, is the solution of P1, ω1 represents the unit cost of the computing power service used by the i-th smart device, ω2 is the weight of the offloading service, Indicates the maximum CPU frequency used by the i-th smart device for local computing.

4. The method according to claim 3, characterized in that: The step 3 comprises: Given a relay selection strategy and uplink spectrum allocation strategy Joint optimization task division strategy Computing resource allocation strategy Two sub-problems are expressed as: The objective function and constraints in subproblem P2 are both convex functions. Convex optimization techniques are used to solve P2 and obtain the local optimal solution under the current number of iterations. Given a task division strategy Computing resource allocation strategy Jointly optimized relay selection strategy and uplink spectrum allocation strategy Two sub-problems are expressed as: in 5. The method according to claim 4, characterized in that The step 4 comprises: Using the slack variable method, the discrete variable x i,j Relax to 0≤x i,j ≤1; make The value of θ is approximately 0; For the second-order terms Introducing auxiliary variables where x i,j and Respectively restricted to 0≤x i,j ≤1 and Convert P3 to: Use convex optimization technology to solve P3.1, get the local optimal solution under the current number of iterations, and substitute the solution into And the optimization variable x is rounded off. i,j Perform recovery.

6. The method according to claim 1, characterized in that The channel access in the two stages from the smart device to the relay device and from the relay device to the base station adopts orthogonal frequency division multiplexing.

7. The method according to any one of claims 1 to 6, characterized in that: The convex optimization method adopted in step 4 is a CVX solver.

8. The method according to claim 1, characterized in that The task processing modes of step 1 include: local execution, relay offloading and edge computing.

9. An electronic device, characterized in that: including memory and processor; The memory is used to store computer programs; The processor is configured to implement the method according to any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images